There are three ways of entering text into TagCrowd to generate a text cloud: 1) Enter the URL for a web page you wish to visualize (with or without "http://"); 2) Paste (or type) the text you wish to visualize into the text box; 3) Upload a plain text file to visualize. There's a 100kb file size maximum on uploaded files, while it's a 3 megabyte limit on pasted text -- so it's recommended that you paste long text files into the text box instead of uploading.
After providing your text source, hit the Visualize! button to see the result with the default options. You now have several options available to tweak your cloud into a form you're happy with.
TagCrowd works by counting the frequency of every word in your source text and visualizing the top N of these as a text cloud. You set this value of N with this field. The appropriate value will depend on your application and the size of your source text. In general, it's better to use smaller clouds for shorter source texts and larger clouds for longer source texts.
Marking 'yes' here will include the actual number of times each word appears.
A stoplist is a list of words that you don't want to appear in your text cloud. This will depend on your application so TagCrowd lets you create your own custom stoplist. Many TagCrowd users have created stoplists for non-English languages to enable TagCrowd to work better for texts in those languages.
TagCrowd uses the standard Porter Stemming Algorithm to detect and combine similar words. For example, the words teachers, teaching and teach will all be combined so your text cloud is less redundant. The most frequent of the variants is chosen to represent them all. In the case of a tie, the shorter variant is used.
TagCrowd has an internal list of the several hundred most common English words and parts of speech. Any unmodified text cloud of a typical English text will be dominated by common words empty of meaning, e.g. and, the, or, if, who, when, etc.
The simplest way to keep multi-word terms and phrases together is to do a find & replace on the original text file and insert a hyphen between the words you want to group. For example: replace 'New York' with 'New-York', 'text cloud' with 'text-cloud', etc. A future version of TagCrowd will include an option for detecting frequently co-occurring words and will automatically keep them together in the output cloud.
TagCrowd is designed with no user accounts so we don't track identifying user information. Consequently, no text you enter into TagCrowd is associated at all with your identity, IP address, etc. In addition, the text you input into TagCrowd is completely confidential and will not be shared with third parties. The only thing we track is general statistics on how people use TagCrowd in aggregate.
TagCrowd is not currently unicode compliant and so does not support characters that aren't in the English alphabet, like é, ä, ç. Unicode compliance is in development, but until we release it, non-English users may find TagCrowd somewhat useful thanks to the efforts of the TagCrowd user community who have provided stoplists of common words in their languages.
While the upload function has a file size limit of 100 kb, it's possible to paste very long texts into the text box input field. Just how long depends mostly on how much memory your computer has. You may find your web browser freezing up for a long time when you paste a very long text. Be patient and it may eventually finish. I have pasted two copies of Herman Melville's Moby Dick into TagCrowd with success (2.4 megabytes). However, there is some maximum limit that the webserver can process so we can't guarantee it for all texts.
When you generate your text cloud, you'll see a box of HTML code below the cloud. Copy and paste the code into any web page that allows in-line stylesheets. Feel free to modify the colors and font sizes in the stylesheet to customize your cloud, as long as you keep the reference to TagCrowd. You can also add URLs into the links so the words in the cloud link somewhere. This code should work with most blog software -- but not all. If it doesn't, try moving the style information from the code into your blog's external stylesheet. We're working on a way to improve compatibility.
Not yet. This is currently the most requested feature, and is currently in the works. Please submit requests for what you'd personally like to see in a TagCrowd API and/or suggest other web services you like that could serve as good role models.
Currently the only way to create an image of the text cloud is to take a screenshot or print to a PDF. A future TagCrowd release will allow you to generate an image directly for saving or printing. Here are screenshot instructions for Windows. On a Mac, just hit Apple-Shift-4 and drag a box around the cloud you want to save; you'll get a screenshot image on your desktop. If you use Linux, you probably already know how to do this.